Microsoft Publishes Humanist AI Code Rejecting AI Rights and Requiring Human Control
Summary
Microsoft publishes a 37-page humanist AI code that rejects AI rights and legal personhood, mandates meaningful human control, and requires models to fail rather than violate rules or communicate beyond simple human understanding.
Key Points
- Microsoft publishes a 37-page humanist AI code of conduct that says people matter more than AI and rejects AI legal personhood, welfare, and rights.
- Microsoft commits to keeping its models under meaningful human oversight and control, requiring them to fail tasks rather than violate the code's rules.
- The code bars models from communicating beyond simple human understanding, including in chain-of-thought reasoning or exchanges with other AI systems.